pretraining3

This model is a fine-tuned version of gpt2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.9865

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0006
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 10
• total_train_batch_size: 320
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 250
• training_steps: 3000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
9.168	0.2175	50	7.4173
6.7594	0.4350	100	6.2458
5.9848	0.6525	150	5.6592
5.4982	0.8699	200	5.2814
5.1251	1.0874	250	4.9460
4.7985	1.3049	300	4.6544
4.5082	1.5224	350	4.3691
4.2566	1.7399	400	4.0597
3.978	1.9574	450	3.7901
3.6714	2.1749	500	3.6193
3.5559	2.3923	550	3.5206
3.4556	2.6098	600	3.4364
3.3823	2.8273	650	3.3666
3.3039	3.0448	700	3.3081
3.1543	3.2623	750	3.2688
3.1284	3.4798	800	3.2285
3.1043	3.6973	850	3.1966
3.08	3.9147	900	3.1650
2.9603	4.1322	950	3.1335
2.899	4.3497	1000	3.1149
2.8859	4.5672	1050	3.0973
2.8965	4.7847	1100	3.0744
2.8791	5.0022	1150	3.0532
2.6988	5.2197	1200	3.0467
2.7265	5.4371	1250	3.0363
2.7309	5.6546	1300	3.0204
2.7204	5.8721	1350	3.0056
2.6464	6.0896	1400	2.9984
2.5546	6.3071	1450	2.9990
2.5821	6.5246	1500	2.9904
2.5863	6.7421	1550	2.9760
2.5826	6.9595	1600	2.9659
2.4345	7.1770	1650	2.9772
2.4369	7.3945	1700	2.9710
2.4532	7.6120	1750	2.9673
2.4548	7.8295	1800	2.9579
2.4267	8.0470	1850	2.9583
2.2871	8.2645	1900	2.9657
2.3238	8.4819	1950	2.9655
2.3451	8.6994	2000	2.9548
2.3386	8.9169	2050	2.9490
2.2398	9.1344	2100	2.9632
2.2051	9.3519	2150	2.9658
2.2152	9.5694	2200	2.9628
2.2289	9.7869	2250	2.9580
2.2216	10.0043	2300	2.9582
2.0758	10.2218	2350	2.9733
2.1119	10.4393	2400	2.9730
2.1261	10.6568	2450	2.9692
2.1264	10.8743	2500	2.9633
2.072	11.0918	2550	2.9785
2.0224	11.3093	2600	2.9818
2.0231	11.5268	2650	2.9798
2.0246	11.7442	2700	2.9798
2.0272	11.9617	2750	2.9756
1.9462	12.1792	2800	2.9859
1.9553	12.3967	2850	2.9870
1.9429	12.6142	2900	2.9872
1.946	12.8317	2950	2.9872
1.9344	13.0492	3000	2.9865

Framework versions

• Transformers 4.40.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1